For things related to energy costs, responsibility is often shuffled between the facilities management team and...
By submitting your personal information, you agree that TechTarget and its partners may contact you regarding relevant content, products and special offers.
IT. But who is ultimately responsible for a well-maintained UPS, which also falls under disaster recovery?
An uninterruptable power supply (UPS) has one simple job: to immediately and seamlessly take over the task of providing power to a data center when the main supply from the grid is interrupted. As such, it has been seen as part of the fabric of the data center facility and has fallen under the auspices of facilities management (FM). After all, the procurement and management of the grid-based energy supply is FM's -- why not that of the UPS?
The issue is that the world within the data center is changing. Energy costs are highly variable but trending inevitably upward. Equipment densities are increasing, with many racks now drawing in excess of 15 kW.
In a 2011 study, Jonathan Koomey, consulting professor at Stanford University's civil and environmental engineering department, calculated that power usage across data centers worldwide had increased by 56 % between 2005 and 2010 -- less than the originally expected 100 %, but still taking up to around 26 GW of energy needed to power data center facilities -- or the output of 26 average-sized power stations.
An average power usage effectiveness (PUE) of 1.8 means that the energy used by non-IT equipment is around 14.5 GW. The majority of that is in cooling down data centers, but a large part is also in running UPS systems.
A typical UPS system consists of a mechanism for providing stored energy, generally through the use of rechargeable batteries. Such systems will be able to provide continuous power for only a matter of minutes or, at best, hours, and so need to be backed up by alternative sources of power generation -- which tend to be petrol- or diesel-powered generators.
Although a UPS is nominally off for most of the time, the energy impact of a poorly implemented or managed system can be massive. Older systems would let a continuous trickle drain from the main grid supply in order to ensure that the batteries were kept fully charged. Overcharging batteries can lower their effective life, so many UPSs had complicated systems to manage the power being drawn; however, these could also be poorly effective and resulted in lower energy efficiencies. In order to optimize the batteries' life, they would need to be deep-drained on a regular basis, requiring either the data center to be run from the UPS until the batteries were sufficiently drained, or shunting the output to a false load -- in each case introducing a weak link so that if the grid power failed, the UPS would have less capability to kick in long enough for the generators to take over.
Additionally, a petrol- or diesel-powered generator requires someone to either: drain and replace the fuel on a regular basis to prevent it from going stale; or check that the system is running on a regular basis so the fuel is used up. With smaller diesel generators being far less energy-effective than centralized main power generators, such fuel cycling can be costly at a financial level and hit an organization's green credentials hard. However, running the data center via the UPS and generator does, at least, provide a test for how effective the power backup procedure is.
Another issue is that as energy densities build up in the data center, the need to introduce more UPS capability to match the growth in data center need also grows. This has driven a move from the monolithic UPSs of old to a more modular style, with a corresponding growth in the use of in-rack and in-row UPSs such that incremental needs can be met as new racks and rows are introduced.
However, at the generator level, this is not so easy. Sourcing incremental capability and ensuring that the generator's output is matched with other generators' output as the need arises can be difficult -- and most data centers choose to go for a forklift upgrade and replace the generators completely.
But tying the data center equipment energy usage with the priorities around the workloads on the equipment can lead to a more intelligent and useful UPS strategy.
It takes both the facilities management team and IT to tango
Although an organization is increasingly dependent on IT, it is not equally dependent on all of the components of its IT. Whereas a retail organization losing power to its e-commerce website could lead to the failure of the company, losing power to servers that support its payroll is not so much of an issue. Using the intelligence built into modern equipment can lead to CPU cores being powered down when not in use, disk drives being spun down and low-energy states used where necessary.
Forming the strategy around which workloads to fully support and how low-energy states can be used to maintain different levels of IT capability is not an FM task. IT has to work with the business in order to prioritize workloads and what levels of capabilities have to be maintained over a period of time. The use of integrated in-rack/in-row UPS systems provides the granularity required for IT to be able to create an optimized power usage strategy. However, the technicalities behind modern UPS systems, such as power factor control and total harmonic distortion of the input current waveform , are best suited for the FM team, which needs to understand how the UPS system will fit in with the broader energy strategy and needs of the overall organization.
As expected, the answer to the question of whether maintaining a UPS is an FM or IT matter is that it is both. FM and IT have to work together very closely to ensure that any system put in place is ready to perform. It must support the organization in maintaining critical IT workloads while also minimizing overheads and energy waste.
ABOUT THE AUTHOR: Clive Longbottom is the cofounder and service director at Quocirca and has been an IT industry analyst for more than 15 years. Trained as a chemical engineer, he worked on anti-cancer drugs, car catalysts and fuel cells before moving on to IT. He has worked on many office automation projects, as well as on Control of Substances Hazardous to Health, document management and knowledge management projects.